The improvement of P management in agriculture and environment requires a good understanding of residual effect of applied P in soils. The specific adsorption of P on variable charge minerals has been considered as the major mechanism that leads to a very low utilization of P fertilizer by crops within a growing season in Chinese red soils. Soil incubation and isotope tracing analysis were carried out to examine the transformation kinetics and potential availability of added specifically sorbed P-32 in two pH contrasting light textured soils. The P-32 recovered by 0.5 M NaHCO3 extraction and microbial biomass-P measurement from the added specifically sorbed P-32 in the soils was well described by a first-order reaction and a Langmuir-type kinetic model, with correlation coefficients (R) being, on average, 0.938 and 0.959, respectively. The half-life (t(1/2), from the first-order model) of the four tested mineral-P complexes ranged from 29 to 47 d in the acid sandy soil and 33 to 105 d in the neutral silty soil. Goethite-P was the most stable among the four tested mineral-P complexes. The potential availability of the mineral complex P (q(m), in percent of total P-32 added) obtained from the Langmuir equation ranged from 43.7 to 90.9% for the four mineral-P complexes, and decreased in the order: Al oxide-P (90.9%) > montmorillonite-P (86.2%) > kaolinite-P (77.5%) > goethite-P (60.2%) in the acid sandy soil, whereas the order was Al oxide-P (89.3%) > kaolinite-P (86.2%) > montmorillonite-P (82.6%)> goethite-P (43.7%) in the neutral silty soil. Based on the release rate and potential availability, kaolinite-P and Al oxide-P could be important sources for residual effect of applied P in variable-charge soils. The goethite-P has the lowest release rate and potential availability among the mineral-P complexes, implying that iron oxides may be the most important variable-charge mineral responsible for P fixation in the Chinese red soils.